基于递归定量分析与多核学习支持向量机的玻璃纤维增强复合材料缺陷识别技术

Defect Identification of Glass Fiber Reinforced Polymer Based on Recurrence Quantitative Analysis and Multiple Kernel Learning SVM

为了提高玻璃纤维增强复合材料(Glass Fiber Reinforced Polymer,GFRP)超声检测中缺陷识别技术的准确性,提出基于递归定量分析(Recurrence Quantitative Analysis,RQA)与多核学习支持向量机(MKLSVM)相结合的检测模型,以提高检测GFRP中不同类型缺陷的能力。结果表明,该模型能够准确识别GFRP中的分层缺陷与夹杂缺陷,检测识别率达到92.92%,并且与基于离散小波变换(Discrete Wavelet Transform,DWT)和经验模态分解(Empirical Mode Decomposition,EMD)的MKLSVM检测模型的识别率相比,所提出的检测模型的识别率分别提高了7.5%和3.75%。

In order to improve the accuracy of defect identification technology in ultrasonic testing of GFRP,a detected model based on the recurrence quantitative analysis(RQA)and multiple kernel learning support vector machine(MKLSVM)was proposed to improve the ability to detect different types of defects in glass fiber reinforced polymer(GFRP).The results show that the proposed detected model can accurately identify delamination defects and inclusion defects in GFRP,and the detection recognition rate reaches 92.92%.Compared with the recognition rate of the MKLSVM detection model based on discrete wavelet transform(DWT)and empirical mode decomposition(EMD),the recognition rate of the proposed detection model has increased by 7.5% and 3.75%,respectively.